DataLearner logo
GE

Gemma 4 26B A4B

Chat modelFunction callingGemmaGemma 4

Google Gemma 4 26B A4B

Release date: 2026-04-02Updated: 2026-08-23Knowledge cutoff: 2025-01Views: 1,300
Live demoGitHubHugging FaceCompare
Parameters
25.2B
Context length
256K
Multilingual
No data
Reasoning ability
No data

Gemma 4 26B A4B is a chat model from DeepMind, released on 2026-04-02. It accepts text and image input and returns text output. The cataloged parameter count is 25.2B, with 3.8B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 32K. Cataloged capabilities include Reasoning model. The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Gemma 4 26B A4B

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Mode (Default)Standard Mode
Context length
256K tokens
Max output length
32K tokens
Model type
Chat model
Modality (in / out)
Text, Image → Text
Release date
2026-04-02
Model file size
No data
MoE architecture
Yes
Total params / Active params
25.2B / 3.8B
Knowledge cutoff
2025-01
Gemma 4 26B A4B

Open source & experience

Code license
Apache 2.0
Weights license
Apache 2.0- Commercial use permitted
GitHub repo
N/A
Hugging Face
huggingface.co/google/gemma-4-26b-a4b
Live demo
ai.google.dev/gemma
Gemma 4 26B A4B

Official resources

Paper
Paper
DataLearnerAI blog
DataLearnerAI blog
Gemma 4 26B A4B

API details

API speed
No data
No public API pricing yet.
Gemma 4 26B A4B

Benchmark Results

Gemma 4 26B A4B currently shows benchmark results led by LiveCodeBench (37 / 127, score 77.10), MMLU Pro (50 / 133, score 82.60), GPQA Diamond (127 / 270, score 82.30). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage
Internet

General Knowledge

3 evaluations
Benchmark / mode
Score
Rank/total
MMLU Pro
Thinking Mode
82.60
50 / 133
HLE
Thinking Mode
8.70
164 / 185
HLE
Thinking ModeToolsInternet
17.20
144 / 185

General Evaluation

1 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Thinking Mode
82.30
127 / 270

Coding and Software Engineer

2 evaluations
Benchmark / mode
Score
Rank/total
CodeForces
Thinking Mode
1718
17 / 20
LiveCodeBench
Thinking Mode
77.10
37 / 127

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
τ²-Bench
Thinking ModeTools
68.20
26 / 43

Math and Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
AIME 2026
Thinking Mode
88.30
17 / 19

Multimodal Understanding

1 evaluations
Benchmark / mode
Score
Rank/total
MathVision
Thinking Mode
82.40
10 / 12

Long Context

1 evaluations
Benchmark / mode
Score
Rank/total
GDM-MRCR v2 (8-needle, 128K)
Thinking Mode
44.10
6 / 8
View benchmark analysisCompare with other models

Compare with other models

Gemma 4 26B A4B

Publisher

Google Gemma 4 26B A4B

Model Overview

Gemma 4 26B A4B is a chat model from DeepMind, released on 2026-04-02.

It accepts text and image input and produces text output. Its cataloged capabilities include Reasoning model. The cataloged parameter count is 25.2B, with 3.8B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 32K.

The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. The evaluation section contains 8 cataloged benchmark results with their recorded modes and scores. The page links 4 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

Gemma 4 26B A4B

FAQ

What is Gemma 4 26B A4B?

Gemma 4 26B A4B is a chat model from DeepMind, released on 2026-04-02. It accepts text and image input and returns text output. The cataloged parameter count is 25.2B, with 3.8B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 32K. Cataloged capabilities include Reasoning model. The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

What input and output modalities does Gemma 4 26B A4B support?

The current model record lists text and image as input and text as output.

What are the main recorded specifications for Gemma 4 26B A4B?

The cataloged parameter count is 25.2B, with 3.8B active parameters per inference. The recorded context window is 256K, and the recorded maximum output is 32K. Fields without a source-backed value remain undisclosed.

Why is no API price shown for Gemma 4 26B A4B?

Open-weight checkpoint; Google publishes no billable first-party token price for this exact checkpoint.

Are benchmark results available for Gemma 4 26B A4B?

The evaluation section contains 8 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is Gemma 4 26B A4B open source?

The checkpoint is listed under the Apache 2.0 license, with the license link included in the model references. Review the linked license text before commercial or derivative use.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code